Hello!
I am running a simply do loop to change the values of a 21*21 double precision real matrix under the RELEASE mode. This loop takes me 58 seconds. (I call cpu_time() at the very beginning of the codes as well as at the end). However, when I comment out the call cpu_time, the running time becomes 87 seconds.
I am surprised to see this large difference in running time (the only difference in the codes is call cpu_time). Anyone has the clue?
It seems that the difference disappears under DEBUG.
The codes that I ran is as follow:
-------------------------------------------------------------------------------------------------------------------------------------
real(kind=8), dimension(21,21) :: OCC
call system_clock(time0)
call CPU_TIME(time0_CPU)
OCC=1
do ic=1,4
do i1=1,90
do i2=1,2000000
OCC=9.D0*9.D0
end do
end do
end do
call CPU_TIME(time1_CPU)
call system_clock(time1, time_rate)
write(*,*) time1_CPU-time0_CPU
write(*,*) real((time1 - time0),8) / real(time_rate,8)
read(*,*)
stop